Regression toward the mean--another form of Gambler's Fallacy

[TwoCatSam]

I feel like blowin' in the wind..............

Is there a regression toward the mean?  Sometimes.  Does red follow black?  Sometimes.

When you run robots and study the trots you will find that when things go wrong, they can stay wrong for eternity.  "The wheel can remain erratic longer than you can stay solvent."

Last week, running the L v F ExcelBot, I saw it go bad.  Since it was play money then, I just let it run to see if it would turn around.  200 Euro later, it had not.  How many real Euro would you truly invest to see if it turned around?  400?  1,000?  Yeah, in the funny papers.

I've seen the SuperRoulette bot run for hours producing twice as many reds as blacks.  I lost 34,000 real pennies over night expecting it to turn around.  It never did.

This bottom line is this:  It MAY regress to the mean and it MAY NOT.  Like everything in roulette, it is not etched in stone.  So don't bet the farm on it.

Sam

[Drazen]

 No Sam.

Regression toward mean is not falacy at all. It is scientificaly provable phenomenon.

In some other thread I explained what it means. But what bothers you here is not that... Unfortunately you felt on your skin actual strenght and full meaning of it... 

Even if you started at 3.0 STD playing continuosly it could stretch long enough for you to be busted. At my time I paid that lesson too...

That was ineventable to happen for you or anyone else playing that way. I knew that will happen and as author of that thread claims that way of betting is more superior then any other 18 vs 18, he is also in big big fallacy. He will attribute loss due to playing on RNG as he doesn't have confidence in it thinking they cheated you. That is not true because BV RNG is the only one RNG you can be sure it produces real not cheating random, becasue it gets it from physical resource and outcomes can't be changed after they generate sequence.

RTM worked perfectly as it should

This destroyed you, not RTM. RTM is harmless, just behaves as it should. No harm feelings from it LoL

If you were using a D'Alembert, it's effectiveness wears off very quickly when the stakes get too high. Increasing the stakes from 2 units to 3 is a 50% increase, but raising from 10 to 11 only gives a 10% increase, and from 20 units to 21 units only a 5% increase - see what I mean? the bigger the stake the less effective is any increase at overcoming any deviation, and if you hit a longer ecart you NEVER catch up. And that is excatly what happened. You wouldnt have chance even if you started at lets say 3.0 and playing conitnuosly, and you, poor guy were rushing on full strength of variance like ruffian.
Like mouse is rushing onto elephant, or even ant is rushing ont elephant?? Oh dear..

Sorry for your loss. Welcome to the world of deviations.

Slightest better chances you would have with this:
Use something like a D'Alembert BUT don't let the stake get so high that it becomes ineffective, if you're losing, cut back the stake to a maximum of one half and either keep it at that while you're still losing or raise by 1 unit after each loss and lower by 1 unit after each win.


Best


Drazen

[TwoCatSam]

Drazen

You have every right to be wrong.

Sam

[Drazen]

Wrong about what Sam?

[ilukan]

When you/and other gents/on this forum grasp what is;LEVELING  D'ALEMBERT PROGRESSION----which I
have writen of in several methods,with all explanations how to addopt it throughout any session......
you may become a winner.Nobody here seem to be interested about it....as they learned classical 1+1-
without any improvements.....This most don't understand.

[Nickmsi]

 Hi Sam . . .

Yes, I believe there is a regression toward the mean (RTM). But the key word here is "TOWARD". 
Eventually, over the long run, there will be a regression TOWARD the mean, it may never achieve the mean but it will move closer to it, or it may actually achieve a balance, but most often it does not.
What I have found interesting is most people assume that if you have a dominance of 60 Blacks and 40 Reds in 100 spins, that the bet should be on RED, as most assume it is "Due". (Gambler's fallacy).
I too thought that way until I tested this theory in one of our bots.  When Black was dominant I would bet on Red, sometimes I would win but most times it would take forever for Red to regress toward the mean and most times it never achieved it.  No flat betting or progression would help.
Now, we have a bot that has tracks several statistics, one of them being dominance.  When we get a 60% Black and 40% Red in 100 spins, we bet the BLACK dominance will continue in the short term.  This bet selection we find more profitable.
Good topic . . .

Nick

[TwoCatSam]

What is the deal?

I wrote several hundred words, posted it, read it and now it's gone.

Who does this stuff and why?

[Skakus]

What is the deal?

I wrote several hundred words, posted it, read it and now it's gone.

Who does this stuff and why?

Type in Word, Sam.

[Skakus]

For betting on EC’s a modified D’Alembert can be a very strong MM.

The thing to remember with RTM is that everything about roulette is a moving target. It never stands still. This includes the number stream used to measure the deviation you are hoping will RTM.

Drazen makes a similar point to me when he says to modify or cap you progression. That’s what I meant by, don’t chase the rabbit too far down the hole.
Conversely you should also include a cap or modification to your D’Alembert for winning periods. Don’t chase the monkey too far up the tree either.

I prefer to play EC’s for dominance to continue over short bursts. RTM can occur at any time over any quantity of spins, or it can disappear for ages, and capping both ends of the betting spectrum keeps you inside the moving target more effectively.

When results move well outside the 3std point, whether winning or losing, that is a good time to make adjustments or apply modifications to the progression. If things remain the same you still win, if things RTM, you still win. Without some mod’s to the progression, if things remain the same you still win, if things RTM, you will lose some or all of what you won and possibly more.

[Turner]

For betting on EC’s a modified D’Alembert can be a very strong MM.

The thing to remember with RTM is that everything about roulette is a moving target. It never stands still. This includes the number stream used to measure the deviation you are hoping will RTM.

Drazen makes a similar point to me when he says to modify or cap you progression. That’s what I meant by, don’t chase the rabbit too far down the hole.
Conversely you should also include a cap or modification to your D’Alembert for winning periods. Don’t chase the monkey too far up the tree either.

I prefer to play EC’s for dominance to continue over short bursts. RTM can occur at any time over any quantity of spins, or it can disappear for ages, and capping both ends of the betting spectrum keeps you inside the moving target more effectively.

When results move well outside the 3std point, whether winning or losing, that is a good time to make adjustments or apply modifications to the progression. If things remain the same you still win, if things RTM, you still win. Without some mod’s to the progression, if things remain the same you still win, if things RTM, you will lose some or all of what you won and possibly more.

Ok, I'm not an expert in this, but I understand all you say...and Drazen. what i do understand is Blackjack..and there is a standard basic stratedgy. if this...do this...if that, do that. it can be memorised, and on first glance, it looks like a cheat sheet on a piece of paper, but it is actually a very complete view of best practice.
In theory, RTM must evoke best practice and basic stratedgy, just like blackjack.

Why do you not see cheat sheets being produced on this subject if you "just wait" until results are moving sround 3STD . This subject is portrayed as mechanical as BJ basic stratedgy.
like Speed used to say....when 3STD this that and the other, simply do this. Simples!!
Like we are all refused membership so some esoteric club.

[TwoCatSam]

Skakus

You have a point.  I should write it in word and the paste it.

Not accusing anyone of anything; my computer was screwey all day.

But I remember reading it.  Made a couple of changes.......

Maybe FLAT is right after all--I have lost me marbles!

I'll re-do it later.

Sam

[iggiv]

Regression toward the mean. No fallacy at all.

link:://en.wikipedia.org/wiki/Regression_toward_the_mean

[Skakus]

Ok, I'm not an expert in this, but I understand all you say...and Drazen. what i do understand is Blackjack..and there is a standard basic stratedgy. if this...do this...if that, do that. it can be memorised, and on first glance, it looks like a cheat sheet on a piece of paper, but it is actually a very complete view of best practice.
In theory, RTM must evoke best practice and basic stratedgy, just like blackjack.

Why do you not see cheat sheets being produced on this subject if you "just wait" until results are moving sround 3STD . This subject is portrayed as mechanical as BJ basic stratedgy.
like Speed used to say....when 3STD this that and the other, simply do this. Simples!!
Like we are all refused membership so some esoteric club.

I’m no expert either. Anything I say is just my opinion.

Blackjack and roulette differ because in blackjack past results are not reintroduced back into the game whereas in roulette they are. This makes blackjack more suitable to mechanical strategies with best practice cheat sheets, etc. But yes, in theory RTM must evoke best practice and basic strategy, just like blackjack.

I don’t know how Speed or anyone else plays, but I don’t wait for 3std’s I bet until the results thus far breach 3std’s or so, then I close that window and open a new window to play on with fresh results as they happen, not looking back to the older results anymore.

Works well enough for me until I find something better.

[TwoCatSam]

 I am no mathematician or statistician.  I don't mean to put on that hat.  I read what iggiv linked. 

Let us take an exaggerample.  Say the wheel produced 100 reds in a row.  Or ten. Or twenty.  Any number.  We have our number and we say the wheel must now regress toward the mean.  Why?  Why is a legitimate question.  Newton asked why when the apple fell and hit him on the head.  (I know it's not true!!)

So we have X number of reds and we feel it will correct itself in the next X spins.  If that is true, something must make red less likely to hit and black more likely to hit.  When each spin is a unique event in the universe, why is this true?

As Normy said, it may regress toward the mean but never get there.  How exactly does that help us?  And is there any guarantee it will even regress?   What if more reds come?

The closest real explanation comes from R.D. Ellison.  He spoke of statistical pressure.  He said the numbers “seem” to want to even out. 

If there were a true regression toward the mean, all we would have to do is program a bot to seek a certain Z score or whatever and begin betting.  We would always win.  Folks, it just ain’t so.  I've tried it.

Lastly, many people confuse what is regressing.  Is it the actual numbers or the percentages?  I won't do the math again, but being 60/40 in 100 spins and 60/40 in a million spins is two totally different things.  While the percentage shrinks, the actual number of R v B increases.  How will you profit from that?

Drazen

When you say “poor guy” the meaning is not lost.  You are saying “poor ignorant Sam”.  “Poor silly Sam.”  “Poor loser Sam.”  So you've won enough for a villa in France, huh?  Since you opened the door to insults, you are one of those people who talk a good talk, but you don't walk an inch.  Why not make some videos or post some screen shots of how much you make?  Take a page out of Ralph's book.

I hear you good people say time and time again you just don't trust the casino.  B.S.  Dublin is as honest as they get.  If you wanted to show us how good you are, you'd be on there letting it all hang out. BV is an honest RNG.  Why not get on there and show your stuff?

Van Cliburn just died. You suppose he ever ranted about how well he could play the piano but never graced a seat with his butt? Nope, he played.

So to anyone who says they can make money from regression toward the mean, I ask you to show us. Otherwiseâ€"as far as real-money gambling is concernedâ€"it’s pure fallacy.

Sam

[Drazen]

 Oh Sam c mon , why are you so offensive and interpret any my statement to you in that direction? I didn't had slightest aim toward that direction... I am not best foreign English speaker but I still do understand some words I use. If you see a pedestrian hit by a car, wouldnt you run over him  to help saying oh poor guy, I hope he will be ok?

When you see starving kids in Africa on TV, don't you say: Oh poor fellows I whish I could help them by getting them some food to eat... You use that words but you feel  emphaty, don't you?

I have seen people who speak native English and they used this phrase excatly in that direction. And I can still say poor ignorant Sam (to have same feeling as above) thinking he is not aware/doesn't understands what got onto him.. It is not an insult. So what is wrong with that word here?

Second of all, I don't have to prove anything to anyone nor I will ever intend to. I am not guru of roulette of any kind and I don't know best and don't want to brag about anything. I am raised to be modest at first place, and like that I behave in roulette world also.

Ok back to our roulette subject.

I have written with big bolded letters one word. Have you seen it? RTM is not causable phenomenon and you obviously want it to be, and getting mad why it isnt. Simply becasue it isnt. It means RTM doesn't casue future spins. A sequence with a strong deviation doesn't CAUSE the following sequence to be closer to the average, but that's what actually happens. The trick is to find the right entry point.

...Regression happens in different spin frames simultaneously and in multiple aspects, but obviously it always tends to be interpreted as some form of fallacy, and that's not what I mean. Another way is saying that events tend to "balance". They do, but the trick is to not rely on a single probability as a guide. By that I mean; don't assume that because red and black will balance out you should always bet for it every time you see an imbalance (e.g. you see 10 blacks in a row, then bet red). You need to track multiple events - calculate the expectation for each event and act on the basis of deviations from it. The more events you have, the better. The accuracy will be improved if you act when multiple events "point" to the same outcomes. If there is conflict, don't make the bet...

What it does though is keep the deviations to an acceptable (lower) level such that is easyer to make something using money management and progressions. This is called diversification and that is also mathematicaly provable to help to reduce variance. Bayes helped me a lot in pointing and understanding pletny of things, to say in the end. I am fan of his in way.

You can read about that here:link:://en.wikipedia.org/wiki/Diversification_%28finance%29 

In other words, there is still no magical triggers and will ever be...What all those deviations do  is just make you bets SLIGHT easyer but still all is on your judgment and evaluating situation. There is no SURE and MAGIC triggers. Deal with it.

If this way of play works for one, doesn't means it will for someone else... Overall it is still enough hard, don't worry. And I didn't said I can win easy, without stress and quickly. Too bad it doesn't go so... 

I am saying all the time, one just has to find its own way.

Best, sir

Drazen